SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2409.01035 · 2024

Task-Specific Directions: Definition, Exploration, and Utilization in Parameter Efficient Fine-Tuning

arXiv · PDF · Open in the Atlas

Code that ran

6 of 10 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 6 failures of our instrument; 4 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2409.01035.json · badge: /api/ran/2409.01035.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
Chongjie-Si/Subspace-Tuning canonical (classifier label)
classifier basis: census
6 ran · 4 no recorded run
FunctionStatusWhere it lives
evaluate Ran
ran; our contract draft was wrong, not the code
Chongjie-Si/Subspace-Tuning/CR_MR/multi_dataset_eval.py
code served (permissive licence) · get_code("2dc17f5e6fb39b3f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import subprocess

torch.manual_seed(0)

# The function takes dataset (string) and gpu (string/int)
# It constructs a shell command and runs it with subprocess.run
# The dataset parameter is used in an f-string for the command
# The gpu parameter is used in CUDA_VISIBLE_DEVICES and in an f-string for output

ARGS = {
    "dataset": "wikitext",
    "gpu": "0"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
generate_prompt Ran
ran; our contract draft was wrong, not the code
Chongjie-Si/Subspace-Tuning/CR_MR/finetune.py
code served (permissive licence) · get_code("56e674752db00649")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# The function expects a dict with 'input', 'instruction', and 'output' keys
data_point = {
    'instruction': 'Translate the following text to French.',
    'input': 'Hello, how are you?',
    'output': 'Bonjour, comment allez-vous?'
}

ARGS = {
    'data_point': data_point
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
generate_prompt Ran
ran; our contract draft was wrong, not the code
Chongjie-Si/Subspace-Tuning/CR_MR/generate.py
code served (permissive licence) · get_code("0dbc7fc6c518d0fb")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

ARGS = {
    'instruction': 'What is the capital of France?',
    'input': 'I need to know about European capitals.'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
generate_prompt Ran
ran; our contract draft was wrong, not the code
Chongjie-Si/Subspace-Tuning/CR_MR/commonsense_evaluate.py
code served (permissive licence) · get_code("1c6b8be890b6118a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# The function generate_prompt takes two parameters:
# - instruction: a string describing a task
# - input: an optional string providing further context

instruction = "Translate the following text from English to French."
input_text = "Hello, how are you?"

ARGS = {
    "instruction": instruction,
    "input": input_text
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
load_data Ran
ran; our contract draft was wrong, not the code
Chongjie-Si/Subspace-Tuning/CR_MR/commonsense_evaluate.py
code served (permissive licence) · get_code("56a9466ff35b35bb")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import json
import os
import tempfile

# Create a temporary directory structure with a test dataset file
temp_dir = tempfile.mkdtemp()
dataset_dir = os.path.join(temp_dir, 'dataset', 'example_dataset')
os.makedirs(dataset_dir, exist_ok=True)

# Create a sample test.json file
test_data = [
    {"id": 1, "text": "sample text 1", "label": 0},
    {"id": 2, "text": "sample text 2", "label": 1}
]

test_file_path = os.path.join(dataset_dir, 'test.json')
with open(test_file_path, 'w') as f:
    json.dump(test_data, f)

# Change to the temporary directory so the relative path works
original_cwd = os.getcwd()
os.chdir(temp_dir)

# Create a simple args object
class Args:
    def __init__(self, dataset):
        self.dataset = dataset

ARGS = {
    'args': Args(dataset='example_dataset')
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
translate_state_dict_key Ran
ran; our contract draft was wrong, not the code
Chongjie-Si/Subspace-Tuning/CR_MR/export_state_dict_checkpoint.py
code served (permissive licence) · get_code("a4f5d0414a1cf9c6")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

# The function takes a single string parameter 'k' (a state dict key)
# and returns a translated key name or None

# Test with various valid state dict keys that the function handles
ARGS = {
    'k': 'base_model.model.model.embed_tokens.weight'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
main Not yet run
unverified (harvested, no recorded run)
Chongjie-Si/Subspace-Tuning/CR_MR/commonsense_evaluate.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from chitar/QuadEnhancer, is pointer only (licence: NONE) · get_code("415f46d8a341e28d")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from chitar/QuadEnhancer, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

main Not yet run
unverified (harvested, no recorded run)
Chongjie-Si/Subspace-Tuning/CR_MR/math_evaluate.py
code served (permissive licence) · get_code("4740c3e235eedfb4")
How it ran

no invocation recorded.

permute Not yet run
unverified (harvested, no recorded run)
Chongjie-Si/Subspace-Tuning/CR_MR/export_state_dict_checkpoint.py
code served (permissive licence) · get_code("a2f49e13f6e13d6b")
How it ran

no invocation recorded.

unpermute Not yet run
unverified (harvested, no recorded run)
Chongjie-Si/Subspace-Tuning/CR_MR/export_state_dict_checkpoint.py
code served (permissive licence) · get_code("6e275f232bbfb9b6")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 distinct functions with an identity, 9 are also present in other papers whose code we hold (9 with at least one paper dated earlier, 0 the same month, 9 later) and 1 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
generate_prompt 15 other papers
6 dated earlier, 8 later, 1 same year, order not determinable
2304.01933 (2023-04)
LLM-Adapters: An Adapter Family for Parameter-Efficient Fine-Tuning of Large Language Models
2307.08072 (2023-07)
Do Emergent Abilities Exist in Quantized Large Language Models: An Empirical Study
2024.naacl-long.132 (2024 (year only))
LLM-based Medical Assistant Personalization with Short-and Long-Term Memory Coordination
and 12 more not named here
translate_state_dict_key 11 other papers
9 dated earlier, 2 later
2023.findings-emnlp.269 (2023 (year only))
An Empirical Study of Instruction-tuning Large Language Models in Chinese
2304.08177 (2023-04)
Efficient and Effective Text Encoding for Chinese LLaMA and Alpaca
2305.00447 (2023-05)
TALLRec: An Effective and Efficient Tuning Framework to Align Large Language Model with Recommendation
and 8 more not named here
unpermute 11 other papers
9 dated earlier, 2 later
2023.findings-emnlp.269 (2023 (year only))
An Empirical Study of Instruction-tuning Large Language Models in Chinese
2304.08177 (2023-04)
Efficient and Effective Text Encoding for Chinese LLaMA and Alpaca
2305.00447 (2023-05)
TALLRec: An Effective and Efficient Tuning Framework to Align Large Language Model with Recommendation
and 8 more not named here
load_data 10 other papers
5 dated earlier, 5 later
2304.01933 (2023-04)
LLM-Adapters: An Adapter Family for Parameter-Efficient Fine-Tuning of Large Language Models
2401.00788 (2024-01)
Astraios: Parameter-Efficient Instruction Tuning Code Large Language Models
2402.09353 (2024-02)
DoRA: Weight-Decomposed Low-Rank Adaptation
and 7 more not named here
generate_prompt 9 other papers
5 dated earlier, 4 later
2304.01933 (2023-04)
LLM-Adapters: An Adapter Family for Parameter-Efficient Fine-Tuning of Large Language Models
2401.00788 (2024-01)
Astraios: Parameter-Efficient Instruction Tuning Code Large Language Models
2402.09353 (2024-02)
DoRA: Weight-Decomposed Low-Rank Adaptation
and 6 more not named here
permute 9 other papers
7 dated earlier, 2 later
2023.findings-emnlp.269 (2023 (year only))
An Empirical Study of Instruction-tuning Large Language Models in Chinese
2305.00447 (2023-05)
TALLRec: An Effective and Efficient Tuning Framework to Align Large Language Model with Recommendation
2306.05064 (2023-06)
K2: A Foundation Language Model for Geoscience Knowledge Understanding and Utilization
and 6 more not named here
generate_prompt 5 other papers
3 dated earlier, 2 later
2304.01933 (2023-04)
LLM-Adapters: An Adapter Family for Parameter-Efficient Fine-Tuning of Large Language Models
2401.00788 (2024-01)
Astraios: Parameter-Efficient Instruction Tuning Code Large Language Models
2402.09353 (2024-02)
DoRA: Weight-Decomposed Low-Rank Adaptation
and 2 more not named here
evaluate 3 other papers
1 dated earlier, 2 later
2402.09353 (2024-02)
DoRA: Weight-Decomposed Low-Rank Adaptation
2510.03276 (2025-10)
QuadEnhancer: Leveraging Quadratic Transformations to Enhance Deep Neural Networks
2605.08110 (2026-05)
BaLoRA: Bayesian Low-Rank Adaptation of Large Scale Models
main 2 other papers
1 dated earlier, 1 later
2402.09353 (2024-02)
DoRA: Weight-Decomposed Low-Rank Adaptation
2510.03276 (2025-10)
QuadEnhancer: Leveraging Quadratic Transformations to Enhance Deep Neural Networks

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Large language models demonstrate impressive performance on downstream tasks, yet they require extensive resource consumption when fully fine-tuning all parameters. To mitigate this, Parameter Efficient Fine-Tuning (PEFT) strategies, such as LoRA, have been developed. In this paper, we delve into the concept of task-specific directions (TSDs), which are critical for transitioning large models from pretrained states to task-specific enhancements in PEFT. We propose a framework to clearly define these directions and explore their properties and practical utilization challenges. We then introduce a novel approach, LoRA-Dash, which aims to maximize the impact of TSDs during the fine-tuning process, thereby enhancing model performance on targeted tasks. Additionally, based on our exploration of TSD, we focus on an important issue in PEFT: the initialization of LoRA. While some works have pointed out the significance of initialization for LoRA's performance and proposed various strategies, these methods are often empirical and not task-specific. To address this issue, we propose LoRA-Init. Starting from TSD, we identify the directions that require the most adjustment during fine-tuning for downstream tasks. By initializing the matrices in LoRA with these directions, LoRA-Init significantly enhances LoRA's performance. Moreover, we can combine LoRA-Dash and LoRA-Init to create the final version of LoRA based on TSDs, which we refer to as LoRA-TSD. Extensive experiments have conclusively demonstrated the effectiveness of these methods, and in-depth analyses further reveal the underlying mechanisms behind their success.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2409.01035")
get_code_for_paper("2409.01035")
have("2409.01035")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2409.01035.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2409.01035.svg)](https://syntology.ai/paper/2409.01035)

Connect an agent — have() is free.